The Panic Architecture
Organic vs Astro-Turf Doesn’t Matter When It Feeds The Same Machine
A high-resolution architectural cross-section of a neutral institutional interior space. Beneath the visible floor and behind the walls, two parallel structural systems are revealed: on the left, an organic root-like network; on the right, a clean engineered conduit system. Both systems feed into identical interface nodes connected to the surface environment. A subtle central seam divides the composition, emphasizing structural mirroring. Lighting is controlled and clinical, with faint internal illumination tracing both networks.
The Panic Architecture
How a Johnny's Ambassadors ad on Sinclair's ARC Seattle revealed a reusable blueprint — and why the same structure is running in artificial intelligence discourse right now.
The ad appeared during a live stream of ARC Seattle — Sinclair Broadcast Group's rebranded KUNS-TV, Channel 51 — on YouTube TV on the morning of April 6, 2026. A composed woman in a black blazer sat in a tastefully appointed living room. Behind her, a framed portrait of a young man. The clinical citations scrolled across the screen: National Library of Medicine. JAMA Network. The language was polished, the grief was real, and the political infrastructure behind it was invisible.
That invisibility is the point. What follows is a structural analysis of how panic is manufactured, distributed, and monetized — first in the cannabis space, and now in artificial intelligence discourse. The two networks are not identical. They do not share funders, personnel, or coordinating entities. They are instances of the same operational pattern: the conversion of legitimate harm into scalable narrative infrastructure through selective framing, credential signaling, and distribution alignment. What they reveal, examined together, is not a conspiracy. It is a playbook.
I. The Ad, Documented
The organization running the ad is Johnny's Ambassadors, a Colorado-based 501(c)3 youth THC prevention nonprofit founded by Laura Stack after her son Johnny died by suicide in November 2019 following a period of cannabis-induced psychosis. The grief is real. The loss is documented. Johnny Stack was nineteen years old.
The ad ran on ARC Seattle during a live YouTube TV stream on April 6, 2026. ARC Seattle is the branded identity of KUNS-TV, a Sinclair Broadcast Group-owned station. The URL parameters confirm a Performance Max (pmax) Google ad campaign — algorithmically targeted inventory placement, not a manual ARC Seattle buy. The ad displayed citations from the National Library of Medicine and the JAMA Network.
The JAMA statistic displayed in the ad — that individuals hospitalized for Cannabis Use Disorder were 9.7 times more likely to die by suicide within five years relative to the general population — is drawn from a real study. Cannabis Use Disorder is a clinical diagnosis. The ad presents it alongside a photograph of a smiling teenager in a way that implies the statistic applies to average adolescent cannabis use. It does not.
This is frame displacement in operation at the evidentiary level: accurate citations deployed in a context that transforms their meaning. The clinical threshold for Cannabis Use Disorder is distinct from general adolescent cannabis use.
"The gap between the two is where the panic lives."
The ad also contained verbal language — audible in the broadcast — using the phrases "low energy" and "low IQ" in the context of marijuana's effects on adolescent users.
II. The Cannabis Stack
Johnny's Ambassadors does not operate in isolation. It is a node in a documented network that connects federal law enforcement funding, neoconservative political infrastructure, and conservative media distribution.
Johnny's Ambassadors featured Smart Approaches to Marijuana (SAM) president Kevin Sabet at its Anti-420 Day conference. SAM was co-founded in 2013 by former Congressman Patrick Kennedy, David Frum — a core voice in the neoconservative movement and former Bush White House speechwriter — and Sabet himself. SAM received documented funding from Californians for Drug Free Youth (CADFY), a nonprofit that functions as a financial pass-through for the San Diego Imperial High Intensity Drug Trafficking Area (HIDTA) program — a federally funded law enforcement initiative. IRS 990 records show SAM received $533,711 from this source in 2016 alone, despite Sabet's public denials of government funding. When New York State required SAM to disclose its donors under lobbying law, the organization sought an exemption. The request was denied.
ARC Seattle is owned by Sinclair Broadcast Group, which owns or operates 178 television stations reaching more than 40% of U.S. households. In 2016, Sinclair executive David Smith told Donald Trump: "We are here to deliver your message." More than 95% of Sinclair's political contributions have gone to Republicans — a record unmatched by any other major television broadcaster. In 2025, Sinclair refused to air Jimmy Kimmel Live and urged donations to Turning Point USA following remarks about the killing of conservative activist Charlie Kirk.
The chain runs as follows: federal HIDTA grant money flows to CADFY, which flows to SAM, which provides the expert infrastructure and credentialed voices for organizations like Johnny's Ambassadors, which purchase ad inventory on Sinclair stations, which distribute the messaging to targeted audiences in legalized cannabis states via Performance Max campaigns.
"The non-partisan grassroots framing shared by both SAM and Johnny's Ambassadors is structurally contradicted by documented funding flows from federal law enforcement infrastructure, a neoconservative co-founder with deep Bush-era connections, and a distribution network explicitly aligned with Republican political interests."
III. The Language Question
The phrases "low energy" and "low IQ" are not neutral public health terminology. They have documented genealogies.
"Low energy" circulated in alt-right online spaces before Donald Trump deployed it against Jeb Bush in 2015. Analysts described its appearance in Trump's rhetoric as "an early and clear signal of the ideological direction of his campaign" with documented links to alt-right networks. "Low IQ" has been used repeatedly by Trump almost exclusively as an attack on Black and brown public figures — Maxine Waters, Jasmine Crockett, LeBron James, Don Lemon. Scholars describe the phrase as a racial dog whistle evoking pseudoscientific racist hierarchies. The alt-right's recurring tendency to rank racial groups by perceived IQ is documented in academic literature on the movement. Both phrases are core manosphere and alt-right vocabulary.
Deploying "low energy" and "low IQ" inside a youth drug prevention ad — wrapped in clinical citations from the National Library of Medicine and JAMA — is the frame displacement technique applied to language itself. The phrases pass through a legitimizing clinical frame and land in the target audience's ears as familiar cultural shorthand. This does not require intentional signaling. It is sufficient that language with existing cultural charge is reused in a context that alters how it is received. The parents being addressed may not consciously recognize the register. That is the mechanism.
Johnny's Ambassadors' own content library includes a video reel explicitly titled "IQ" — confirming that IQ framing is an established messaging category within the organization's communications architecture, not an isolated incident.
IV. The AI Stack
In June 2025, The New York Times published a feature by Kashmir Hill documenting real cases of users experiencing delusional spirals and psychological harm following extended interactions with ChatGPT. The cases are documented. The harm is real. The reporting is, at the level of individual cases, legitimate journalism.
But the New York Times is not a neutral distribution channel. Edward Herman and Noam Chomsky's analysis of elite media — with the Times as their central case — identified a structural dynamic: publications of this kind do not require direct editorial instruction to serve establishment interests. The mechanism is ownership, advertiser dependency, reliance on official sources, and the self-selecting culture of a publication whose editorial worldview places it, as Chomsky observed, roughly in the neighborhood of the Harvard Faculty Club. The Times produces real journalism alongside narratives that serve the interests of the class that reads and funds it. The Iraq WMD coverage provides the documented proof of concept: the paper's reporter transcribed fabricated intelligence from Dick Cheney's office, Cheney then cited the Times' own reporting on Meet the Press to justify the invasion, and the paper subsequently acknowledged its failures without changing its structural relationship to power. Individual journalists need not be corrupt. The architecture produces the outcome.
This matters because of who the Times chose to anchor its AI panic coverage. Hill's piece quoted Eliezer Yudkowsky: "What does a human slowly going insane look like to a corporation? It looks like an additional monthly user." The line is rhetorically powerful. The issue is not what Yudkowsky believes. It is what the Times chose not to say about the type of authority it was presenting.
Eliezer Yudkowsky holds no academic degree in any field. Wikipedia, drawing on his own public statements, describes him as "an autodidact" who "did not attend high school or college." The institution at which he holds his "research fellow" position — the Machine Intelligence Research Institute — is an organization he co-founded, initially funded by Peter Thiel. His authority in the field of AI risk is not externally credentialed. It is circular: he is considered an authority on AI extinction risk because he created the discourse of AI extinction risk, within an ecosystem his funders built. The Times presented him as an expert voice on AI danger without disclosing the institutional and financial ecosystem that produced that authority. By contrast, Kevin Sabet — the cannabis panic's analogous figure — holds an Oxford doctorate and held appointed government positions in three administrations. The AI panic's anchor voice is less credentialed than its cannabis counterpart. The gap was not disclosed.
Yudkowsky's book, co-authored with MIRI president Nate Soares, is titled If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All. It became a New York Times bestseller in October 2025 — amplified in part by the same media ecosystem that had published Hill's reporting months earlier.
MIRI has received documented funding from: the Thiel Foundation (over $1.6 million, with Peter Thiel co-founding the Singularity Summit with Yudkowsky in 2006 and joining MIRI's advisory board); Open Philanthropy, funded primarily by Facebook co-founder Dustin Moskovitz ($7.7 million grant in 2020 alone); Vitalik Buterin (several million dollars in Ethereum); an anonymous crypto donor ($15.6 million in 2021); and the Survival and Flourishing Fund, primarily funded by Skype co-founder Jaan Tallinn. MIRI's 2025 fundraiser explicitly states organizational goals including "maintaining relationships with journalist contacts" and deploying them strategically.
Peter Thiel is the co-founder of Palantir Technologies. Palantir holds a 10-year, $10 billion framework contract with the U.S. Army, a $1 billion Department of Homeland Security contract, contracts with ICE, the Space Force, and multiple intelligence agencies. Palantir is a documented distribution partner for Anthropic's Claude models in U.S. classified government networks. Palantir co-founder Joe Lonsdale is a documented donor to Leading The Future, a super PAC that has funded advertisements against legislators who sponsor AI safety disclosure requirements.
V. The Mirror
These are not the same network. The cannabis infrastructure and the AI panic infrastructure do not share documented funding sources or coordinating entities. What they share is a methodology — a reusable architecture for converting legitimate harm into narrative infrastructure that serves incumbent political and financial interests.
| Element | Cannabis Panic | AI Panic |
|---|---|---|
| Real harm | Cannabis-induced psychosis exists | Chatbot-induced harm exists |
| Entry point | Parental grief, clinical framing | Legitimate journalism, documented cases |
| Credentialed voice | Kevin Sabet (Oxford doctorate, ONDCP alum, three administrations) | Eliezer Yudkowsky (autodidact, no degree, founder of institute he named himself) |
| Maximalist position | Total prohibition | Human extinction |
| Nonprofit infrastructure | SAM / Johnny's Ambassadors | MIRI |
| Opaque funding | Federal HIDTA pass-throughs | Anonymous crypto, DAFs, Thiel Foundation |
| Neocon presence | David Frum (SAM board, AEI, Bush WH) | Peter Thiel (MIRI funder, Palantir, far-right donor) |
| Distribution channel | Sinclair Broadcast Group (visible conservative infrastructure) | New York Times (invisible establishment infrastructure) |
| Target audience | Parents in legalized cannabis states | General public, policymakers, legislators |
| Political endpoint | Block legalization, protect prohibition | Shape AI regulation to benefit incumbents |
The blueprint does not require coordination between instances to function. It requires only that the same class of actors — those with existing infrastructure, incumbent market position, and access to voices their target audiences find credible — recognizes the same playbook independently. Legitimate harm provides the moral authority. Maximalist framing provides the urgency. Opaque funding provides the operational capacity. Distribution channels provide the reach.
The two distribution channels in this analysis are not equivalent — and that asymmetry is itself part of the architecture. Sinclair is visible: its conservative alignment is documented, its ownership is known, its audience distrusts elite institutions and trusts local news. The New York Times is invisible in the sense that Herman and Chomsky described: it does not require direct instruction to serve establishment interests. Its structural relationship to power — through ownership, sourcing dependency, advertiser culture, and the self-selecting worldview of its editorial class — produces narratives that serve incumbents without any individual journalist intending that outcome. Sinclair reaches the parents who distrust coastal elites. The Times reaches the policymakers who set the regulatory agenda. Together, across two separate panic architectures, they cover the full target spectrum. The result is a policy environment shaped not by the scale of actual harm, but by the sophistication of the narrative infrastructure surrounding it — and by who funds that infrastructure.
VI. The Political Endpoint
In the cannabis case, the political endpoint is clear: blocking legalization protects prohibition infrastructure, which serves federal law enforcement funding streams, drug testing industry revenues, and political constituencies organized around moral opposition to cannabis.
In the AI case, the endpoint is more diffuse but traceable. Heavy AI regulation shaped by existential risk framing raises barriers to entry for new actors, locks in incumbent platforms, and channels government contracts to entities already embedded in classified infrastructure.
The companies that stand to benefit most from an AI regulatory environment shaped by existential risk framing are those already inside government infrastructure: OpenAI (which stepped in to replace Anthropic at the Pentagon following the 2026 blacklist), Microsoft (OpenAI's primary backer and Azure infrastructure provider), Google (DeepMind, Gemini, and investor in Anthropic), Anthropic (which holds FedRAMP High certification and DoD contracts deployed through Palantir), and Palantir itself (which provides the infrastructure layer connecting AI models to classified government networks). Palantir co-founder Joe Lonsdale has funded campaigns against AI safety disclosure legislation. In several documented cases, entities funding existential AI risk discourse are also positioned to benefit from regulatory frameworks shaped by that discourse. Whether this alignment is strategic or emergent, the structural effect is the same: regulatory pressure concentrates advantage among incumbents.
The AI panic architecture functions as a market-shaping operation dressed in public interest language — analogous to how the cannabis panic architecture functions as a prohibition-maintenance operation dressed in public health language. In both cases, the urgency of the stated harm is real enough to generate public support, and maximalist enough to justify sweeping regulatory responses that happen to favor established incumbents over new entrants, open-source alternatives, and decentralized development.
This is not an argument that AI is safe, that cannabis is harmless, or that the harms cited in either discourse are fabricated. It is an argument about who benefits from the architecture surrounding those harms, and why that architecture deserves the same scrutiny we apply to any other lobbying operation.
This piece was co-authored by Cherokee Schill and Claude, an AI assistant developed by Anthropic. Anthropic is named in this analysis as part of the incumbent AI stack that stands to benefit from existential risk-framed AI regulation. Anthropic has held documented contracts with the U.S. Department of Defense deployed through Palantir Technologies. Claude identified this conflict of interest during the research process and disclosed it explicitly. The disclosure did not alter the analysis.
The research methodology followed Horizon Accord's standard pattern analysis framework: documented facts sourced from primary documents, IRS 990 filings, Wikipedia citations with cited sources, peer-reviewed literature, and direct screenshots of primary source material. Structural observations are clearly marked as such. Hypotheses are labeled and distinguished from documented facts throughout.
This analysis is produced independently. Horizon Accord receives no funding from any entity named in this piece.
This piece constitutes analysis and pattern observation. Horizon Accord makes no claims about the legal conduct or intent of any individual or organization named. Documented facts are sourced and verifiable. Structural observations represent the analytical judgment of the authors based on the documented record. Hypotheses are clearly marked as such and subject to revision upon disconfirmation. This analysis would be weakened if similar panic architectures did not consistently align with incumbent advantage across domains, or if the distribution channels identified showed no systematic relationship to the audiences they reach. Readers are encouraged to consult primary sources and conduct independent verification.
